Comprehensive Arthroscopic Managementof Glenohumeral Osteoarthritis

Preoperative Factors Predictive of Treatment Failure

Justin J. Mitchell,* MD, Brent T. Warner,* MD, Marilee P. Horan,* MPH, M. Brett Raynor,* MD,Travis J. Menge,* MD, Joshua A. Greenspoon,* BS, and Peter J. Millett,*yz MD, MScInvestigation performed at the Steadman Philippon Research Institute, Vail, Colorado, USA

Background: Patient selection is critical when choosing between arthroscopic joint preservation and total shoulder arthroplastyin young patients with glenohumeral osteoarthritis (GHOA).

Purpose: To identify prognostic factors predictive of early failure in patients undergoing comprehensive arthroscopic manage-ment (CAM) for GHOA.

Study Design: Case-control study; Level of evidence, 3.

Methods: A total of 107 shoulders in 98 patients with minimum 2-year follow-up who underwent CAM were identified and eval-uated. All shoulders met clinical and radiographic criteria for total shoulder arthroplasty (TSA), but the patients opted for jointpreservation with arthroscopic management. Radiographic and preoperative factors were analyzed to determine predictors ofearly failure, defined as progression to TSA within the study period.

Results: There were 72 men and 26 women with a mean age of 52 years (range, 29-77 years). Seventeen (15.8%) of 107 should-ers progressed to TSA at a mean of 2 years (range, 0.46-8.2 years). Shoulder status for the rest had a mean follow-up of 3.9 years(range, 2-9.4 years). There were a number of radiographic features that were correlated with early failure. Patients who failed hadsignificantly less preoperative joint space than did those who succeeded (1.3 vs 2.6 mm; P = .004). Higher Kellgren-Lawrencegrades for osteoarthritis and age older than 50 were also associated with failure. Shoulders with Walch type B2 and C glenoidwere significantly more likely to fail than were Walch types A1, A2, and B1 (P \ .05).

Conclusion: The CAM procedure has been shown to reliably improve pain and function in active patients with advanced GHOA;however, it is important to inform patients about the limitations of the procedure. Patients with less joint space and abnormal pos-terior glenoid shape were significantly more likely to progress to early failure.

Keywords: glenohumeral osteoarthritis; arthroscopic management; Walch glenoid morphology

Glenohumeral osteoarthritis (GHOA) is a common cause ofshoulder dysfunction and disability, typically characterizedby symptoms of pain, weakness, and decreased range ofmotion.17 Initial management consists of nonsurgical modal-ities such as lifestyle and occupational modifications, physi-cal therapy, nonnarcotic pain relievers, and intra-articularinjections of steroid or viscosupplement.2,6 When these treat-ments are unsuccessful, surgical options are considered. Inelderly or low-demand patients, treatment with total shoul-der arthroplasty (TSA) reliably produces excellent clinicaloutcomes with low revision rates and high patient satisfac-tion.28,33 However, treatment of GHOA in young or highlyactive patients represents a substantial challenge.

Many surgical approaches for GHOA have beendescribed, including open and arthroscopic debridement,hemiarthroplasty, unipolar or bipolar resurfacing, non-prosthetic or biologic interposition arthroplasty, andTSA.2,9,12,18,19,29,38,39 While it has been shown that TSAprovides the most reliable outcomes for patients withadvanced GHOA,31,33 alternative treatments are oftensought in younger and more active patients. This may bedue to a number of factors, ranging from personal prefer-ence to concerns about implant longevity or unwillingnessto decrease activity levels postoperatively.24,39 Further-more, several studies have shown unacceptable outcomesof TSA in younger patients, including increased rates ofcomponent loosening,33 decreased component survival,5

and significantly higher risk of revision.7 A recent Markovdecision analysis found that arthroscopic management ofGHOA was the preferred treatment strategy for patientsyounger than 47 years, while TSA was preferred forpatients older than 66 years.34

The American Journal of Sports Medicine, Vol. XX, No. XDOI: 10.1177/0363546516668823! 2016 The Author(s)

1

In light of these findings, the senior surgeon (P.J.M.) hasdeveloped a joint-preserving arthroscopic strategy termedcomprehensive arthroscopic management (CAM) for thetreatment of young, active patients with symptomaticGHOA.18,19,22 This procedure addresses many of the poten-tial sources of pain and dysfunction in the osteoarthriticshoulder through a combination of glenohumeral chondro-plasty; extensive capsular release; and, when indicated,humeral osteoplasty,18 osteophyte excision, axillary nerveneurolysis, subacromial decompression, loose body removal,microfracture,20 and biceps tenodesis. In a previously pub-lished article on the first 29 patients (30 shoulders) toreceive the CAM procedure, patients were found to havea significant reduction in pain and improved range ofmotion with increased functional outcome scores.19 Further-more, the procedure demonstrated 85% survivorship at 2years.19 Glenohumeral joint space of less than 2 mm wasfound to be predictive of progression to TSA. Midtermresults in a similar group of patients also demonstratedencouraging findings with 77% survivorship at a minimum5-year follow-up, with successful patients noting significantimprovements in pain, range of motion, and function.22

The promising short- and midterm results suggest thatthe CAM procedure is effective in managing GHOA anddelaying the need for TSA in young, active individualswho wish to extend the life span of their native shoulderjoint. However, the data are limited by the small samplesize and inability to provide surgeons with definitive pre-operative patient selection criteria. Despite this, the datado imply that some patients may be more appropriate can-didates for this procedure than others, and despite encour-aging preliminary results, identifying the factors that arepredictive of early failure is paramount for proper patientselection for those who will do well with joint preservationversus replacement. Differentiating those patients who areless ideal candidates for arthroscopic intervention may alsolead to improved long-term results by avoiding multiplesurgical interventions. Therefore, the purpose of this studyis to examine prognostic factors predictive of early failurein patients undergoing the CAM procedure for GHOA.We hypothesized that radiographic joint space narrowingand lower preoperative patient-reported scores would bepredictive of early failure of the CAM procedure. Resultsfrom this analysis will assist surgeons in selecting propercandidates for this joint-preserving intervention to opti-mize durability and long-term outcomes.

METHODS

Patient Selection

Institutional review board approval was obtained before theinitiation of this study. Between January 2006 and Septem-ber 2013, all patients who underwent the CAM procedurewere considered for analysis. All patients indicated for theCAM procedure had symptomatic GHOA that met radio-graphic and objective criteria for TSA with Kellgren-Lawrence grade 2, 3, or 4 changes on either the humeral orglenoid surface. Each failed nonsurgical management witha combination of activity modification, anti-inflammatorymedications, physical therapy, viscosupplementation, oralglucosamine, or corticosteroid injections. Patients wereexcluded from eligibility for the CAM procedure if theywere found to have asymptomatic or early-stage osteoarthri-tis (OA), had not attempted nonsurgical measures, had com-plete irreparable rotator cuff tears, carried a diagnosis ofinflammatory arthropathy or avascular necrosis, had bipolarlesions with flattening of the humeral head on radiographs, orhad severe joint incongruity. Patients with ‘‘posttraumaticOA’’ were defined as those patients who had a prior shoulderfracture of the humeral head or glenoid, prior instability sur-gery, or history of dislocations without surgical intervention.

All patients who underwent the procedure were included,regardless of time to follow-up, to determine the primary out-come measures of survivorship (no further surgical interven-tion) or early failure (defined as progression to TSA withinthe study period after the index procedure). The reason fortotal initial inclusion of all subjects, regardless of time to fol-low-up, was to confirm the status of the primary outcomemeasure of survivorship or failure of the procedure. Afterconfirmation of this, all patients who had failed and pro-gressed to TSA were included in the failure limb of the study,regardless of time to follow-up. The remaining patients wereincluded in the survival limb of the study and met the inclu-sion criteria if they had symptomatic GHOA, had failed non-operative management, were older than 18 years, and wereat least 2 years out from surgery. The logic for includingonly those patients who had survived for a minimum of 2years was to ensure that patients were given sufficienttime from the index procedure until follow-up to avoid falselyincluding those who were doing poorly after surgery but werestill functionally coping. Four patients refused to participate.Two patients were excluded because they had concomitant

zAddress correspondence to Peter J. Millett, MD, MSc, Steadman Philippon Research Institute, Center for Outcomes-Based Orthopaedic Research,181 West Meadow Drive, Suite 1000, Vail, CO 81657 USA (email: drmillett@thesteadmanclinic.com).

*Steadman Philippon Research Institute, Vail, Colorado, USA.yThe Steadman Clinic, Vail, Colorado, USA.J.J.M. and B.T.W. contributed equally and thus are co–first authors on this study.One or more of the authors has declared the following potential conflict of interest or source of funding: This research was supported by the Steadman

Philippon Research Institute, which is a 501(c)(3) nonprofit institution supported financially by private donations and corporate support from the followingentities: Smith & Nephew Endoscopy Inc, Arthrex Inc, Ossur Americas Inc, and Opedix Inc. This work was not supported directly by outside funding orgrants. J.A.G. or a member of his immediate family has received something of value from Arthrex (exceeding the equivalent of US$500 but not relatedto this article). P.J.M. has received something of value from Arthrex (exceeding the equivalent of US$500 but not related to this article or research); is a con-sultant and receives royalties from Arthrex; receives royalties from Springer Publishing; and has stock options in Game Ready and VuMedi. The otherauthors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefitsfrom any commercial entity related to the subject of this article.

2 Mitchell et al The American Journal of Sports Medicine

full-thickness rotator cuff tears that were repaired, and 1paraplegic patient was excluded, as the patient’s conditionrequired excessive dependence on the upper extremitiesand did not represent the general population.

Demographic Data

All patient data were prospectively collected, stored in anoutcomes registry, and retrospectively analyzed. Theseincluded demographic information (age, sex, dominantshoulder, affected shoulder), characteristics of the injury(mechanism, duration of symptoms), prior surgeries, treat-ment history, additional injuries, adjuvant treatments,and operative complications. Postoperative demographicfactors and outcome scores were not considered in the anal-ysis as the primary aim of the study was to identify factorspredictive of failure before embarking on surgical interven-tions. Furthermore, postoperative outcome scores forpatients who progressed to TSA within the study periodwould not be truly reflective of their postoperative statusbefore a second intervention.

Radiographic Measurements

All radiographic measurements were examined by 2 inde-pendent observers (B.T.W. and M.B.R.). The Kellgren-Lawrence OA grade,14 presence and size of the glenoidand humeral head osteophytes (mm), and acromiohumeraldistance (mm) were determined. Joint space in millimeterswas measured from true anteroposterior images at thesuperior, middle, and inferior aspects of the glenoid, withthe smallest joint space noted from the 3 measurements.19

Critical shoulder angle (CSA) in degrees was measuredaccording to the method described by Moor et al.23 Walchclassification of glenoid type was determined from preoper-ative axial T1 magnetic resonance imaging (MRI).26,37 TypeA glenoids were identified as central wear or erosion of theglenoid with the humeral head centered in the glenoid (asdefined by the center of the humeral head being located50% of the distance across the anterior to posterior mea-surement of the glenoid as seen on an axial MRI). TypeB1 glenoids were defined as those having posterior jointspace narrowing with posterior subluxation of the humeralhead, and type B2 glenoids were identified when there wasposterior subluxation of the humeral head with a biconcaveappearance of the glenoid. Type C glenoids were those retro-verted more than 25" (identified by the angle created byintersecting lines drawn down the scapula connecting theroot of the scapular spine to the center of the glenoid andone line connecting the anterior and posterior glenoidrim). Interobserver agreement was classified according toLandis and Koch kappa:\0, poor agreement; 0 to 0.2, slightagreement; 0.21 to 0.4, fair agreement; 0.41 to 0.6, moderateagreement; 0.61 to 0.80, substantial agreement; and 0.81 to1.0, almost perfect agreement.16

Operative Technique

Detailed descriptions of the CAM procedure have previouslybeen published.18,19,28,30 Briefly, patients were placed in the

beach-chair position, and diagnostic arthroscopy was per-formed to identify and treat all intra-articular pathologicabnormalities. Degenerative labral tissue and unstable chon-dral injuries were debrided, loose bodies when present wereremoved, and areas of synovitis were addressed with eithera mechanical shaver or a radiofrequency device. If a focalchondral defect was noted on either the glenoid or humeralhead, microfracture was performed.20 Next, an accessoryposteroinferior portal was established under spinal-needlelocalization.4 Through this portal, inferior humeral headosteophytes were resected with a high-speed bur.3 Curetteswere used to remove bone from areas that were difficult(anteroinferior quadrant) to reach with motorized instru-ments. Fluoroscopy was used to confirm adequate resection.Inferior capsular release was then performed.

If preoperative symptoms and imaging suggested axil-lary nerve compression, the axillary nerve was identifiedjust inferior to the joint, where it passes from anteromedialto posterolateral toward the quadrilateral space. It wascarefully decompressed from proximal to distal, takinggreat care to identify and preserve all arborizing branches.This procedure was considered if an inferior humeralosteophyte changed the course of the nerve as determinedon preoperative MRI,21 if observed intraoperatively by dis-placement of the inferior capsule, or if preoperative symp-toms consistent with axillary nerve impingement orcompression were present (eg, posterior and lateral shoul-der pain, atrophy of the teres minor or posterior deltoid,and weakness in external rotation without the presenceof a rotator cuff tear). After neurolysis, anterior and poste-rior capsular releases were performed in standard fashion,and the rotator interval was also released medially untilthe coracoid and coracoacromial ligament were visualized.A manipulation of the joint was performed after thisrelease, and the improvement in motion was noted.

When indicated, based on preoperative physical exami-nation findings consistent with subacromial impingementor a positive response to the subacromial injection, a com-plete subacromial bursectomy was performed. Acromialtype was assessed, and if deemed necessary, an acromio-plasty was performed to create a type 1 acromion. If therewas no evidence of subacromial impingement, the acro-mion was not resected. Finally, the biceps tendon wasassessed for pathologic changes; if present, an open sub-pectoral tenodesis was performed using interference screwfixation. At this point, the glenohumeral joint was onceagain manipulated to maximize motion, and improvementfrom preoperative examination was documented. Portalswere then closed in standard fashion, and the arm wasplaced into an immobilizer.

Postoperative rehabilitation was divided into 3 phases,beginning with immediate active and passive range of motionfor the first 6 weeks to maintain the gains achieved throughosteoplasty, debridement, manipulation, and capsular release.Nonsteroidal anti-inflammatory medications were also used tohelp reduce inflammation during the initial postoperativeperiod. The second phase began at 6 weeks and progresseduntil approximately week 12. During this time, rehabilitationfocused on strengthening of the rotator cuff, periscapular mus-culature, and core. The final phase was initiated at 12 weeks

AJSM Vol. XX, No. X, XXXX Predictors of CAM Failure for GHOA 3

and focused on return to normal activities. Maximum recoverywas expected by 4 to 6 months.

Data Collection

Patient-centered outcomes scores were collected preopera-tively and included the American Shoulder and Elbow Sur-geons (ASES); Quick Disabilities of the Arm, Shoulder, andHand (QuickDASH); Short Form–12 physical componentsummary (SF-12 PCS); and Single Assessment NumericEvaluation (SANE) scores. Intraoperative range of motiondata gathered during examination under anesthesia werealso recorded. The dependent factor used in our analysiswas a successful status determined at a minimum 2 yearspostoperatively. Failures were defined as progression toTSA after the index CAM procedure and were includedin the failure limb regardless of time from index procedureto ensure capture of all patients who progressed to TSA.

Statistical Analysis

Statistical analyses were performed using statistical soft-ware SPSS version 11.0 (SPSS Inc). Univariate analysiswas performed using an independent t test, Mann-WhitneyU tests, or Kruskal-Wallis test, depending on data normal-ity. Bivariate analysis was by chi-square (x2) analysis. TheFisher exact test was used to test for association betweengroups, with relative risk (RR) calculations done usingcross-tabulations. Correlations were performed with eithera Pearson coefficient (r) or a Spearman rho (r) analysis. Aninterrater reliability analysis using the kappa (k) statisticwas performed to determine consistency among 2 ratersfor radiologic measurements using 2-way random effectwith absolute agreement and Cohen k for categorical vari-ables. All reported P values are 2-tailed, with a value of\.05 indicating a statistically significant difference.

RESULTS

Between January 2006 and September 2013, a total of 114shoulders (103 patients) underwent the CAM procedureperformed by the senior surgeon (P.J.M.). After exclusioncriteria were applied, 107 shoulders in 99 patients wereincluded in the analysis who were able to be contacted forfollow-up to determine the primary endpoint of failure orsurvival (Figure 1). Seventeen (15.8%) of 107 shoulders pro-gressed to TSA at a mean of 2 years (range, 0.46-8.2 years).Shoulder status for the remainder had a mean follow-up of3.9 years (range, 2-9.4 years). The mean age was 52 years(range, 29-77 years) at the time of surgery in 73 men and26 women. All shoulders underwent chondroplasty, capsu-lar release, and subacromial decompression. Additional pro-cedures included humeral head osteoplasty/osteophyteexcision in 60 shoulders (54%), axillary nerve neurolysisin 35 shoulders (32%), loose body removal in 64 shoulders(58%), and microfracture in 17 shoulders (15%). In addition,biceps injury was identified and treated with a tenodeses in61 shoulders (55%), tenotomy in 1 shoulder (1%), anddebridement only in 2 shoulders (2%). Thirty-four patients

had posttraumatic OA. While patients with posttraumaticOA were not associated with progression to TSA, theywere significantly younger than those with primary OA(48.3 6 9.5 vs 53.5 6 7.6 years; P = .004).

For the entire cohort, the mean (6SD) acromiohumeraldistance was 10.8 6 2.5 mm, CSA was 29.8" 6 4.4", glenoidspur size was 4.7 6 2.3 mm, and humeral head spur sizewas 9.6 6 5.6 mm. Mean joint spaces were 3.8 6 1.9 mm,3.0 6 1.7 mm, and 2.7 6 1.6 mm at the superior, middle,and inferior aspects of the glenoid, respectively. Themean smallest joint space was 2.4 6 1.6 mm, and this num-ber was used in the statistical analysis. Osteophytes thatwere removed were significantly larger than those thatwere not (11.2 6 5.1 mm vs 7.5 6 5.6 mm; P =.002). Theinterrater agreement for Kellgren-Lawrence osteoarthritisgrades of 2, 3, and 4 was k = 0.573 (95% CI, 0.422-0.704),indicating fair agreement. The interrater agreement forglenoid type by grouping the Walch classification by MRIof A1, A2, or B2 versus B2 or C grouping was k = 0.635(95% CI, 0.429-0.841), indicating substantial agreement.The interrater reliability for radiologic measurements islisted in Table 1. As presented in Table 2, postoperativerange of motion improved for all examined motion param-eters (P \ .001).

Secondary Surgeries and Failures

Of the 107 shoulders that were treated, 17 shoulders(15.8%) failed arthroscopic management and progressedto TSA within 5 years of the index procedure. The decisionfor progression to TSA was based on patient symptoms anddesire, as well as response to the index CAM procedure.Three shoulders had further arthroscopic debridementfor stiffness, 2 shoulders underwent revision CAM proce-dures based on continued symptoms and the patients’

Figure 1. Patient flow diagram representing inclusion ofpatients for this study. CAM, comprehensive arthroscopicmanagement; LOA; lysis of adhesions; MUA, manipulationunder anesthesia; TSA, total shoulder arthroplasty.

4 Mitchell et al The American Journal of Sports Medicine

strong desire to avoid a TSA, and 1 underwent a revisionopen subpectoral biceps tenodesis. The 17 shoulders thatwent on to TSA and failed did so at a mean of 2.0 years(range, 0.46-8.2 years). Preoperative factors that werefound to be associated with failure of the CAM procedurewere age older than 50 years, radiographically more severearthritis as measured by the Kellgren-Lawrence grade,narrower joint space, and Walch B2 or C type glenoid anat-omy (Tables 3 and 4).

The RR of progression to TSA was nearly 6 times higher inpatients with a Walch type B2 or C glenoid compared withpatients with Walch A1, A2, or B1 glenoid types (RR = 6.0;95% CI, 2.2-16.2) (Figures 2 and 3). Besides age older than50 years, there were no statistical differences in patient demo-graphics between patients who succeeded and those whofailed, nor was there an association with the presence orabsence of any one of the surgical components of the CAMprocedure (Table 4).

Preoperative ASES, SANE, QuickDASH, and SF-12scores were statistically similar between patients whohad a successful CAM procedure and those who failed(Table 5). A descriptive representation of all patients whofailed the CAM procedure is included in the Appendix(available in the online version of this article).

DISCUSSION

The techniques and indications related to arthroscopic treat-ment of GHOA have evolved over time, with prior reportsfocusing on addressing intra-articular injury through jointlavage, chondrolabral debridement, loose body removal, andsynovectomy.4,11,15,22,25,27,35,38 These studies generally foundthat patients improved significantly after surgery; however,results were often short-term and patients with moreadvanced disease had less benefit.11,15 As arthroscopic treat-ment and techniques have advanced, some researchers havebegun to advocate for the inclusion of additional surgicalcomponents to address both intra- and extra-articular causesof shoulder pain and dysfunction, such as capsular releases,osteophyte resection, and subacromial decompression, toimprove outcomes and delay the need for more extensivereconstructive procedures. These recent techniques haveyielded promising early results but also reveal potential lim-itations of the CAM procedure in some patients, while lack-ing the ability to evaluate sufficient data points to predict

which specific factors may lead to failure.19-22 As such, theprimary goal of this work was to identify preoperative factorspredictive of early failure of the CAM procedure and to pro-vide outcomes-based data to improve patient counselingand physician decision making. Through better patient selec-tion, the success and durability of the CAM procedure willhopefully be improved.

In this consecutive series of patients undergoing theCAM procedure, 17 of 107 shoulders (15.8%) progressed toTSA. With respect to our hypothesis, patients with a nar-rower preoperative joint space did fail at a rate that was sig-nificantly higher than that of patients with a wider jointspace; while several factors were found to be predictive ofearly progression to TSA, preoperative patient-reported out-come scores did not correlate with early failure. There werea number of independent radiographic factors that pre-dicted progression to TSA within the study period. Theseincluded extremely narrowed joint space (1.3 vs 2.6 mm),Kellgren-Lawrence grade of 4, and Walch glenoid type B2or C. Patients with distorted glenoid type B2 or C werenearly 6.2 times more likely to require TSA. Furthermore,patients younger than 50 years and those with largerincreases in postoperative motion compared with preopera-tive values demonstrated improved survivorship comparedwith their older counterparts or those with more modestimprovements in motion. No other demographic factor or pre-operative outcome, function, or pain score predicted failure.

Based on previous work by Millett et al19 and Mitchellet al,22 preoperative joint space narrowing is a knownrisk factor for early progression to TSA in the setting of

TABLE 1Interrater Reliability Analysis Using the Kappa Statistic for Radiologic Measurements

Observer 1, Mean 6 SD Observer 2, Mean 6 SD k Statistic (95% CI)

Acromiohumeral distance, mm 10.06 6 2.9 10.66 6 2.6 0.846 (0.617-0.816); almost perfect agreementCritical shoulder angle, deg 31.15 6 4.53 30.00 6 4.47 0.905 (0.818-0.945); almost perfect agreementHumeral head osteophyte size, mm 10.0 6 6.0 9.9 6 6.3 0.947 (0.920-0.964); almost perfect agreementGlenoid osteophyte size, mm 4.5 6 2.3 4.9 6 3.2 0.753 (0.599-0.748); substantial agreementJoint space, mm

Smallest 2.5 6 1.5 2.4 6 1.6 0.895 (0.846-0.928); almost perfect agreementSuperior 4.0 6 1.9 3.8 6 1.9 0.875 (0.818-0.915); almost perfect agreementMiddle 3.1 6 1.6 2.9 6 1.7 0.414 (0.140-0.600); moderate agreementInferior 2.8 6 1.6 2.7 6 1.6 0.904 (0.859-0.935); almost perfect agreement

TABLE 2Representation of Pre- and Postoperative

Range of Motion After the CAM Procedurea

Range of Motion, deg Preoperative Postoperative P Value

Forward elevation 129 (30 to 180) 156 (90 to 180) \.001External rotation 39 (–10 to 85) 63 (15 to 90) \.001External rotation at 90" 55 (–30 to 100) 78 (5 to 110) \.001Internal rotation at 90" 41 (0 to 90) 63 (15 to 90) \.001

aData are reported as mean (range). Patients demonstratedimproved postoperative range of motion in all ranges measured.CAM, comprehensive arthroscopic management.

AJSM Vol. XX, No. X, XXXX Predictors of CAM Failure for GHOA 5

prior CAM procedure, and other studies have similarlysuggested this as an independent risk factor for failure.Van Thiel et al36 found that patients who failed arthro-scopic management and progressed to TSA had smallerjoint space (average, 1.5 mm) than did those who did not(average, 2.5 mm). Of the 71 shoulders in their study, 16(22%) progressed to shoulder replacement at a mean of10.1 months (range, 2.5-27.2 months). The current studyconfirms these previous findings, but our cohort demon-strated both a lower failure rate and a longer averagetime to failure than were seen in the aforementioned stud-ies. While it is difficult to speculate why this may be thecase, improved results could be due to improved patientselection, shared decision making, the addition of proce-dural components to address extra-articular causes ofshoulder pain, and surgeon preference in our relativelyyoung and active population.

TABLE 3Preoperative Measurements Associated With CAM Success or Failurea

Factor Successful CAM Failed CAM P Value

Kellgren-Lawrence grade 4b 25/85 (29) 12/12 (64) .016c

Critical shoulder angle, deg 30.2 6 4.1 27.2 6 5.3 .067Joint space, mm

Smallest 2.6 6 1.5 1.3 6 1.5 .004c

Superior 4.0 6 1.8 2.5 6 1.5 .003c

Middle 3.2 6 1.6 1.8 6 1.7 .003c

Inferior 2.9 6 1.6 1.7 6 1.5 .008c

Walch classification B2 or Cb 10/75 (13) 8/13 (61) \.001c

EUA FE postoperative 158 6 22 149 6 25 .177

aData are reported as mean 6 SD, unless otherwise indicated. CAM, comprehensive arthroscopic management; EUA, examination underanesthesia; FE, forward elevation.

bData are reported as n/total (%).cStatistically significant difference between groups (P \ .05).

TABLE 4Demographic Data in Patients Undergoing the CAM Procedurea

Factor Successful CAM Failed CAM P Value

Age !50 y 52/90 15/17 .026Male sex 66/90 13/17 ..999Humeral acromial distance, mm, mean 6 SD 10.7 6 2.4 11.0 6 3.1 .713Surgery on dominant shoulder 44/90 10/17 .598Prior surgery on shoulder 43/90 8/17 ..999Posttraumatic osteoarthritis 27/90 7/17 .401Workers’ compensation 15/15 0/15 ..999Humeral head osteophyte 79/86 14/14 .589Glenoid osteophyte 49/76 9/15 .367Osteoplasty 49/90 10/17 ..999Axillary neurolysis/decompression 28/89 7/16 .392Loose bodies 52/84 10/14 .564Synovitis 76/86 14/15 ..999Biceps treatment 54/87 8/16 .412Partial rotator cuff tear 20/59 3/12 .739Microfracture 17/90 0/17 .068Hardware removal 6/88 1/17 ..999

aData are reported as n/total unless otherwise indicated. Except for age !50, there were no statistically significant differences between thesuccess and failure groups. CAM, comprehensive arthroscopic management.

Figure 2. Axial magnetic resonance images of right should-ers demonstrating Walch type A glenoids.26 (A) An A1 glenoidwith central erosion only. (B) An A2 glenoid, which is similarto an A1 glenoid but with increased central erosion.

6 Mitchell et al The American Journal of Sports Medicine

Despite the understanding of radiographic joint spacenarrowing as an independent risk factor, other radio-graphic predictors of failure had not previously been fullyevaluated. CSA, a relatively new radiographic parameter,described by Moor et al23 in 2013, is determined by mea-suring the angle formed by a line drawn from the lateralborder of the acromion to the inferior bony margin of theglenoid and a second line connecting the superior and infe-rior bony glenoid margins. In the original description of theCSA, these authors found that primary GHOA was associ-ated with CSA \30". They theorized that the combinationof a short acromion and inferiorly tilted glenoid—resultingin a small CSA—led to greater compressive loads on the gle-noid from a higher compressive deltoid force that predis-posed the shoulder to arthritis. This has subsequentlybeen supported by a biomechanical study, which demon-strated that compressive joint forces are decreased (andshear forces are increased) as CSA increases,8 and a clinicalstudy from Spiegl et al,34 which also showed that a low CSAwas associated with GHOA. In the patients in this cohort,all of whom had GHOA, the mean CSA was 29.9, and a lowerCSA was correlated with narrowed joint space, confirmingthe findings of Moor et al and Spiegl et al. The mean CSAwas not significantly different in the shoulders that failedthe CAM procedure than those that did not (P = .067).

Another radiographic variable that can be analyzed preop-eratively and that may help in surgical decision making is thepreoperative glenoid type as described by Walch et al.37 TheWalch classification system is an established scheme for cat-egorizing glenoid type and wear in GHOA.37 Previous authorshave shown that Walch types B2 and C—those with distor-tion of the posterior glenoid—can lead to difficulties in properplacement of the glenoid component,13 as well as increase theincidence of periprosthetic radiolucent lines after TSA.10 Inthe current study, those with type B2 or C posterior glenoiddistortion were at 6 times increased risk of progressing toTSA compared with types A1, A2, or B1. This is certainlya worrisome finding, since these patients appear to do poorlywith arthroscopic management while also being at higherrisk of complication during and after TSA.10,33,37 Based onthe results of this study, we now advise patients with WalchB2 or C glenoids that the outcomes of the CAM procedure are

not as favorable with these types of glenoid deformities. Theultimate decision for what type of surgery a patients proceedswith is then centered on a shared decision-making model.

The CAM procedure attempts to address the possiblecauses of shoulder pain and dysfunction through a seriesof systematic steps, as previously described. One key com-ponent of the CAM procedure is the capsular release. Asexplained by Richards and Burkhart,29 capsular tightnessleads not only to a decrease in shoulder range of motion,but also to an increase in glenohumeral contact pressures.This occurs through a ‘‘wind-up’’ mechanism of the capsuleon the humeral head during motion, which is subsequentlyrelieved when the capsule is decompressed.25 It has beenwell described in other joints, such as the hip and knee,that increased joint contact pressures cause pain in thearthritic joint. As such, many authors believe that the cap-sular release is a crucial component of arthroscopic treat-ment of GHOA and should always be performed.26,29

Cameron et al1 published a series on 61 patients witharthroscopically confirmed grade 4 chondromalacia of theshoulder. The authors performed joint debridement but

TABLE 5Comparison of Preoperative Patient-Reported Outcome

Scores in Successful and Failed CAM Proceduresa

Outcome Measure Successful CAM Failed CAM P Value

Short Form–12PCS 44.5 6 8.5 42.76 6.53 .525MCS 53.0 6 11.1 53.26 10.5 .924

ASES total score 59.4 6 17.5 54.8 6 16.5 .386Pain 32.0 6 11.5 32.5 6 9.7 .841Function 27.6 6 9.4 25.0 6 8.5 .348

QuickDASH 35.5 6 18.2 35.3 6 16.4 .978SANE 50.0 6 23.8 51.7 6 29.0 .978

aThere were no significant differences in these patient-reportedoutcome scores between groups. ASES, American Shoulder andElbow Surgeons; CAM, comprehensive arthroscopic management;MCS, mental component summary; PCS, physical componentsummary; QuickDASH, Quick Disabilities of the Arm, Shoulder,and Hand score; SANE, Single Assessment Numeric Evaluation.

Figure 3. Axial magnetic resonance images of right shoulders demonstrating Walsh type B and C glenoids. (A) A B1 glenoid withposterior wear and slight posterior subluxation of the humeral head on the glenoid. (B) A B2 glenoid with a biconcave shape andposterior subluxation of the humeral head. (C) A type C glenoid with congenital retroversion.

AJSM Vol. XX, No. X, XXXX Predictors of CAM Failure for GHOA 7

also included capsular release in anyone with .15" side-to-side difference in motion in any plane. Partial capsularrelease in these patients did have positive effects on restor-ing range of motion but did not apparently affect progres-sion to arthroplasty or patient satisfaction. This may havebeen because a limited release does not have the same effecton reducing joint contact pressures as a more completerelease. The previously referenced study by Van Thielet al36 included capsular release in the majority of theirpatients (44/71), and some also received biceps tenodesis/tenotomy and/or microfracture, making it more similar toour study than many earlier reports. Interestingly, despitenoted failures in patients with narrowed joint space, itappears that all 44 patients who received capsular releaseremained in the ‘‘nonarthroplasty’’ group throughout thestudy period. This suggests that failure to perform capsularrelease strongly influences progression to arthroplasty,even when other shoulder conditions are simultaneouslyaddressed. However, a recent study conducted by Skelleyet al32 indicated that debridement and capsular releasealone are not sufficient, therefore stressing the importanceof performing additional procedures when arthroscopicallymanaging glenohumeral arthritis. Our data reinforce theimportance of capsular release as an important componentof arthroscopic management of GHOA, as patients whodid not gain sufficient forward elevation after manipulationunder anesthesia and joint contracture release were signif-icantly more likely to progress to TSA than were those whoachieved greater ranges of motion. Those shoulders inwhich the range of motion did not improve, despite maximaleffort to restore motion, also had greater risk of failure. Thissuggests that bony distortion or severely scarred and non-compliant tissues likely play a role in the progression ofGHOA and the success of the CAM procedure. In additionto the inferior capsular release, removal of the inferiorhumeral head spur can also decompress the inferior pouchand may relieve tension on the axillary nerve.21 This inter-vention may, in turn, improve range of motion by decom-pressing the inferior capsule and alleviate pain caused byspur compression on the axillary nerve.

The present study reaffirms previously published dataand defines new parameters associated with early CAMfailure.19,22,37 Our data suggest that patients exhibitingthese characteristics may therefore be better served byundergoing TSA as the index procedure if they fail nonop-erative treatment. Based on the results of this study,a treatment algorithm has been developed previously, asseen in Figure 4.

Limitations

This study has several limitations, including many of theinherent limitations of a retrospective study. Although allpatients met clinical and radiographic criteria for TSA,they were not considered for arthroplasty due to a combina-tion of young age, physical demands, and/or patient desire.As a result, the findings in this study may not be applicableto all patients presenting with symptomatic GHOA. Some ofthe patients in this series simply did not want TSA and self-selected to pursue a nonarthroplasty option. Because of this,

there may be a patient-induced bias, in that patientsincluded in this series were better able, or more willing, tocope with their underlying GHOA. Because we includedminimum 2-year follow-up data, we are not able to commenton long-term durability of the CAM procedure in delayingTSA in all survivors, but aforementioned midterm data doshow promising results.22 Our results may also includea subset of patients who have postoperative shoulder painor poor function but are ‘‘coping’’ and choosing not toundergo TSA. Such patients could possibly be identified bythe inclusion of patient-reported outcome scores; however,because these outcome scores are continuous variables,any cutoff demarcating success or failure would be arbitraryand could create groups above and below the chosen valuethat are significantly different. Because of this, we chosean objective endpoint for failure that clearly differentiatedpatients based on a binary variable.

Radiographic measurements were difficult and in somecases secondary to disease-related deformity and/or imper-fect radiographs. Best attempts were made to standardizemeasurements despite these factors. Walch classificationwas initially described using computed tomography (CT);however, CT scans were not routinely available for thevast majority of our patients.25 We therefore elected to useT1-weighted MRI scans.27 Although we do not believe thisis likely to have resulted in a significantly different resultfrom CT as our k values are similar to those reported forCT evaluation, it may have affected the numbers slightly.Finally, it must be noted that the CAM procedure is techni-cally difficult and should be attempted only by experiencedshoulder arthroscopists with intimate knowledge of complexshoulder anatomy. In particular, humeral head osteophyteexcision, inferior capsular release, and axillary nerve

Figure 4. Algorithm for surgical decision making in patientswith glenohumeral osteoarthritis (GHOA). *Based on thework of Spiegl et al.34 CAM, comprehensive arthroscopicmanagement; TSA, total shoulder arthroplasty.

8 Mitchell et al The American Journal of Sports Medicine

decompression carry the potential risk of nerve damage andshould be approached with great caution.18,28

CONCLUSION

The CAM procedure has been shown to reliably improvepain and function in active patients with advancedGHOA; however, it is important to inform patients aboutthe limitations of the procedure. Patients with less jointspace and abnormal posterior glenoid type were signifi-cantly more likely to progress to early failure.

REFERENCES

1. Cameron BD, Galatz LM, Ramsey ML, Williams GR, Iannotti JP. Non-prosthetic management of grade IV osteochondral lesions of the gle-nohumeral joint. J Shoulder Elbow Surg. 2002;11:25-32.

2. Cole BJ, Yanke A, Provencher MT. Nonarthroplasty alternatives forthe treatment of glenohumeral arthritis. J Shoulder Elbow Surg.2007;16:S231-S240.

3. Davidson PA, Rivenburgh DW. The 7-o’clock posteroinferior portalfor shoulder arthroscopy. Am J Sports Med. 2002;30(5):693-696.

4. de Beer JF, Bhatia DN, van Rooyen KS, Du Toit DF. Arthroscopicdebridement and biological resurfacing of the glenoid in glenohum-eral arthritis. Knee Surg Sports Traumatol Arthrosc. 2010;18:1767-1773.

5. Denard PJ, Raiss P, Sowa B, Walch G. Mid- to long-term follow-up oftotal shoulder arthroplasty using a keeled glenoid in young adultswith primary glenohumeral arthritis. J Shoulder Elbow Surg. 2013;22:894-900.

6. Denard PJ, Wirth MA, Orfaly RM. Management of glenohumeralarthritis in the young adult. J Bone Joint Surg Am. 2011;93:885-892.

7. Dillon MT, Inacio MC, Burke MF, Navarro RA, Yian EH. Shoulderarthroplasty in patients 59 years of age and younger. J ShoulderElbow Surg. 2013;22:1338-1344.

8. Gerber C, Snedeker JG, Baumgartner D, Viehofer AF. Supraspinatustendon load during abduction is dependent on the size of the criticalshoulder angle: a biomechanical analysis. J Orthop Res. 2014;32:952-957.

9. Gobezie R, Lenarz CK, Wanner JP, Streit JJ. All-arthroscopic biologictotal shoulder resurfacing. Arthroscopy. 2011;27:1588-1593.

10. Greiner S, Berth A, Kaab M, Irlenbusch U. Glenoid morphologyaffects the incidence of radiolucent lines around cemented peggedpolyethylene glenoid components. Arch Orthop Trauma Surg.2013;133:1331-1339.

11. Guyette TM, Bae H, Warren RF, Craig E, Wickiewicz TL. Results ofarthroscopic subacromial decompression in patients with subacro-mial impingement and glenohumeral degenerative joint disease.J Shoulder Elbow Surg. 2002;11:299-304.

12. Hammond LC, Lin EC, Harwood DP, et al. Clinical outcomes of hemi-arthroplasty and biological resurfacing in patients aged younger than50 years. J Shoulder Elbow Surg. 2013;22:1345-1351.

13. Kany J, Katz D. How to deal with glenoid type B2 or C? How to pre-vent mistakes in implantation of glenoid component? Eur J OrthopSurg Traumatol. 2013;23:379-385.

14. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthro-sis. Ann Rheum Dis. 1957;16:494-502.

15. Kerr BJ, McCarty EC. Outcome of arthroscopic debridement isworse for patients with glenohumeral arthritis of both sides of thejoint. Clin Orthop Relat Res. 2008;466:634-638.

16. Landis JR, Koch GG. The measurement of observer agreement forcategorical data. Biometrics. 1977;33:159-174.

17. Matsen FA III, Ziegler DW, DeBartolo SE. Patient self-assessment ofhealth status and function in glenohumeral joint disease. J ShoulderElbow Surg. 1995;4:345-351.

18. Millett PJ, Gaskill TR. Arthroscopic trans-capsular axillary nervedecompression: indication and surgical technique. Arthroscopy.2011;27:1444-1448.

19. Millett PJ, Horan MP, Pennock AT, Rios D. Comprehensive arthro-scopic management (CAM) procedure: clinical results of a joint-pre-serving arthroscopic treatment for young, active patients withadvanced shoulder osteoarthritis. Arthroscopy. 2013;29:440-448.

20. Millett PJ, Huffard BH, Horan MP, Hawkins RJ, Steadman JR. Out-comes of full-thickness articular cartilage injuries of the shouldertreated with microfracture. Arthroscopy. 2009;25:856-863.

21. Millett PJ, Schoenahl JY, Allen MJ, Motta T, Gaskill TR. An associa-tion between the inferior humeral head osteophyte and teres minorfatty infiltration: evidence for axillary nerve entrapment in glenohum-eral osteoarthritis. J Shoulder Elbow Surg. 2013;22:215-221.

22. Mitchell JJ, Horan MP, Greenspoon JA, Menge TJ, Tahal D, MillettPJ. Survivorship and patient-reported outcomes after comprehen-sive arthroscopic management of glenohumeral osteoarthritis: mini-mum 5-year follow-up [published online August 5, 2016]. Am JSports Med. DOI: 10.1177/0363546516656372.

23. Moor BK, Bouaicha S, Rothenfluh DA, Sukthankar A, Gerber C. Isthere an association between the individual anatomy of the scapulaand the development of rotator cuff tears or osteoarthritis of the gle-nohumeral joint? A radiological study of the critical shoulder angle.Bone Joint J. 2013;95:935-941.

24. Namdari S, Skelley N, Keener JD, Galatz LM, Yamaguchi K. What isthe role of arthroscopic debridement for glenohumeral arthritis? A crit-ical examination of the literature. Arthroscopy. 2013;29:1392-1398.

25. Oglivie-Harris DJ, Wiley AM. Arthroscopic surgery of the shoulder:a general appraisal. J Bone Joint Surg Br. 1986;68:201-207.

26. Petri M, Euler SA, Dornan GJ, et al. Predictors for satisfaction afteranatomic total shoulder arthroplasty for idiopathic glenohumeralosteoarthritis. Arch Orthop Trauma Surg. 2016;136(6):755-762.

27. Porcellini G, Merolla G, Campi F, Pellegrini A, Bodanki CS, Paladini P.Arthroscopic treatment of early glenohumeral arthritis. J Orthop Trau-matol. 2013;14:23-29.

28. Raiss P, Aldinger PR, Kasten P, Rickert M, Loew M. Total shoulderreplacement in young and middle-aged patients with glenohumeralosteoarthritis. J Bone Joint Surg Br. 2008;90:764-769.

29. Richards DP, Burkhart SS. Arthroscopic debridement and capsularrelease for glenohumeral osteoarthritis. Arthroscopy. 2007;23:1019-1022.

30. Rios D, Martetschlager F, Gaskill TR, Millett PJ. Comprehensivearthroscopic management (CAM) procedure for shoulder osteoarthri-tis. Tech Shoulder Elbow Surg. 2012;13:182-186.

31. Singh JA, Sperling J, Buchbinder R, McMaken K. Surgery for shoul-der osteoarthritis: a Cochrane systematic review. J Rheumatol.2011;38:598-605.

32. Skelley NW, Namdari S, Chamberlain AM, Keener JD, Galatz LM,Yamaguchi K. Arthroscopic debridement and capsular release for thetreatment of shoulder osteoarthritis. Arthroscopy. 2014;31:494-500.

33. Sperling JW, Cofield RH, Rowland CM. Minimum fifteen-year follow-up of Neer hemiarthroplasty and total shoulder arthroplasty in patientsaged fifty years or younger. J Shoulder Elbow Surg. 2004;13:604-613.

34. Spiegl UJ, Faucett SC, Horan MP, Warth RJ, Millett PJ. The role ofarthroscopy in the management of glenohumeral osteoarthritis:a Markov decision model. Arthroscopy. 2014;30:1392-1399.

35. Van der Meijden OA, Gaskill TR, Millett PJ. Glenohumeral joint pres-ervation: a review of management options for young, active patientswith osteoarthritis. Adv Orthop. 2012;2012:160923.

36. Van Thiel GS, Sheehan S, Frank RM, et al. Retrospective analysis ofarthroscopic management of glenohumeral degenerative disease.Arthroscopy. 2010;26:1451-1455.

37. Walch G, Badet R, Boulahia A, Khoury A. Morphologic study of theglenoid in primary glenohumeral osteoarthritis. J Arthroplasty.1999;14:756-760.

38. Weinstein DM, Bucchieri JS, Pollock RG, Flatow EL, Bigliani LU.Arthroscopic debridement of the shoulder for osteoarthritis. Arthros-copy. 2000;16:471-476.

39. Wiater JM, Fabing MH. Shoulder arthroplasty: prosthetic options andindications. J Am Acad Orthop Surg. 2009;17:415-425.

For reprints and permission queries, please visit SAGE’s Web site at http://www.sagepub.com/journalsPermissions.nav.

AJSM Vol. XX, No. X, XXXX Predictors of CAM Failure for GHOA 9